Solving The App Discovery Problem

App Discovery

When we launched?Powerslyde?in March 2013, our primary objective was to make it easy for consumers to find and share apps with the people they trusted most, friends and family.

We created a network that allowed individuals to connect to the people they trusted the most and provided a shortcut for our users to get to the app store and get the apps referred to them by family and friends.

The app leaned heavily on UI/UX, so from the user viewpoint, it wasn’t overly complex and was straightforward and easy to understand.

We removed Powerslyde from the iTunes Store back in 2015. Powerslyde is still available in Google Play and is long overdue for an update. With all the changes to the Android libraries and updates to other SDKs, it most likely does not function as intended. Currently, a code review is underway to see what it would take to update the powerslyde to function with the updated libraries.

Pop Culture

We were in Startup Alley at Disrupt in early fall of 2013. An unexpected benefit of our participation was the opportunity that came from it. The appearance of our company and product, Powerslyde, featured in the TechCrunch Disrupt Episodes during Season 1 of Silicon Valley.

Privacy in 2013

Our business was built with privacy at its core. From the first day, we required users to do a several things related to privacy:

1. Decide if they wanted to use Facebook to login or their email and a password.
2. Create a profile
3. Decide if they wanted a “private” or “public” profile. A private profile could not be viewed unless access was granted to the users.
4. Opt-in to the collection of the apps on their phone with the ability to make some of their apps private and therefore unviewable by others.

When we conducted Paid Advertising we calculated the opt-in rate of users on a daily basis. For the first time, we’re sharing some of that information. Interestingly, we are seeing some striking patterns to the privacy environment of 2022.

Over time we studied this further and found additional information that shed light on users based on their login method and other activities.

Interestingly, we found facebook users were more likely to have a public profile. This was attributed to the fact that facebook users were more accustomed to sharing information online.

We found users who logged in with an email and password were 1.9x more likely to have a private account than facebook users.

From the prior chart, it’s not convincing if operating system is as influential. From the graph, looking at registration and login methods, it’s more likely a user that chose facebook to login is less concerned with privacy than the users who login with email and password.

When thinking about app discovery,?who has the most to gain, and lose?

In the earliest days of the app stores, a developer could be relatively assured of a large number of organic downloads, by just being in the app stores. In 2013 that wasn’t true, and it’s no longer true today.

As we dug deeper into the app discovery problem we discovered an article by?Ouriel Ohayon?where he discussed app discovery as a four-sided issue. The issues he outlined in 2012, are still true today:

• Developer?—?“A mobile app is a business,”?and should be managed accordingly.
• App Stores?—?“They are only stores. They are not marketing or advertising companies.”?Like any other online retailer they do little more than provide a place to distribute your apps directly to consumers. The App Stores, especially Apple, do not let developers control most of the discovery process. While there have been attempts to improve this process over the years, it has actually gotten worse.
• Consumer?—?“Consumers have a hard time finding good apps. People need trusted sources to make quick decisions…”
• Ad Network(s)?—?“Millions of users find out about apps [through] advertising or paid discovery…Paid app discovery is very hard and (to be successful) very expensive.”

These four issues can also be considered the main participants that comprise the foundation of the mobile app economy, and similar to a tectonic landscape, there have been recent privacy changes that have caused shifts to occur and usher in significant change.

Paid UA, an Early Form of App Discovery also known as “Spray and Pray”

In the absence of a solution, the market will attempt to create a workable solution to solve their problem. In the earliest days of mobile advertising, a large number of companies focused their sales approach on convincing an app developer to use “burst” spending to get to the top of the rankings to assist with “organic” discovery. This approach typically required a large volume of ad spend. In effect, this was nothing more than gaming the App Store Rankings. Generate a large number of installs through burst campaigns, waiting for the algorithms to take over to feature your app in the top lists, so that users could organically find your app in store.

This approach was constantly pitched to us at nearly every mobile event we attended. Back in those days, it seemed as if there was a mobile event on a weekly basis, if not more often.

Living on an Island

In order to get our app approved and into the store, we had to remove all third party SDKs and could not send data to any server other than those we owned/managed. This made attribution impossible and presented a real challenge when we wanted to begin paid UA.

It was simple enough to create a link system and collect the minimum amount of data so we could create our system to attribute campaign spend back to each advertiser. It required that we create a fairly sophisticated system to connect everything and get the information we needed. It was no easy task, but once we were finished it made the UA process much simpler, and we were able to complete this without relying on the IDFA.

After completing this system, we were able to capture a more complete view of our users and began to spend a great deal of time understanding the data we were collecting, applying analytics and gathering insights.

App Discovery Insights:

Powerlsyde?was focused on app discovery and was designed so people could give and receive referrals from people they trusted. The primary UI/UX goal was to make the app as simple and easy to use as possible, so that ultimately both the consumer and developer would benefit. One early insight was to find the conversion rate from recommendations by users:

We found the conversion rate was 77%

Other insights we observed:

• People would rather use a free app over a paid app for similar purposes.
• People are highly likely to trust their friends recommendations
• Individuals are focused on saving time where possible
• The less familiar the app the more likely the user is to download it

Business Model

After conducting a great deal of research prior to launch, we found that “of all the apps that consumers download”, on average, 10.89% of those apps are “paid” apps. One of the key assumptions of our business model, was that consumers would continue to download the average 10.89% “paid” applications. Because our business model was predicated on the affiliate program through the app store and we would receive a “commission” whenever someone purchased a “paid” app this was an important factor.

• In reality the download rate was 3.6%, one-third of what was expected.

Paid Advertising works.?It is complex and can be expensive. We had no problem acquiring users at a fairly low price. This is mostly true today if you know what you are doing.

Because the primary monetization model was the affiliate model, we determined it was not sustainable to engage in paid UA on a long term basis.

Had we understood the importance of calculating LTV (Lifetime Value) or ROAS (Return on Ad Spend) at that time, it might have influenced our approach to growing the user base.

Here are the averages for each month we used paid advertising in 2013.

Average Opt-in Rates — 2013

Average CPI (Cost of Install) — 2013

Average Daily Affiliate Revenue — 2013

Average D30 Retention — 2013

In aggregate, our experience in 2013 was no different than the experience of companies in 2022, following the implementation of Apple’s Privacy Rules. Apple’s Privacy Rules have effectively sent the app marketplace backwards to the place it evolved from back in 2013. In many instances, our prior experience could provide significant benefit and insight.

However, the issues today are far worse than they have ever been because of the built up dependence on the IDFA. What could be more valuable than the ability to provide near perfect campaign attribution and downstream event matching. Unfortunately, that was then, and this is now. All forms of Paid UA that were legitimate sources of App Discovery have gotten worse, not better. The only real viable methods today are search and contextual advertising.?The App Discovery Problem has gotten worse, not better.

How can a new system that no longer requires a personal device identifier be created in an efficient way with the least amount of disruption?

As we continued to analyze the daily movements of the app economy from the viewpoint of the apps that reside on a user’s device we continued to gather additional insights..

Notable Privacy Shifts

App Stores:

June 22, 2020 — Apple announced their new?Privacy Rules. These have been updated a number of times to remove ambiguity. The rules were written in such a way that those most educated in the marketplace could easily disagree on the interpretation of a Unfortunately, even among those who have been following and working to educate the marketplace, (Thanks Eric!).

Feb 16, 2022-?Google Introduces the Privacy Sandbox on Android. According to Google, over “90% of the apps on Google Play are free” and “Digital advertising plays a key role in making this possible…That’s why we originally developed advertising ID to give users more control…[We’re] announcing a multi-year initiative to build the?Privacy Sandbox?on Android…” to build “more private advertising solutions”.

July 29, 2022 —?Apple Introducing New App Store Ads to Help Users Discover More Apps. It would be difficult to argue that Apple hasn’t changed their position on marketing and advertising given recent announcements regarding their intention to build/acquire a DSP. This move places Apple with a significant advantage.

“Apple’s ad network utilizes app install and in-app purchase data, to which Apple has exclusive first-party access under the restrictions of ATT, to target ads to users with its ad network. It’s worth underscoring that, with ATT, the scope and substance of consumer data utilized to target ads remains unchanged, except that only Apple has access to it. “

Developer:

To provide the proper context and learn more about the rules impacting developers we simply need to turn to one of the earliest?articles that explains this topic in plain english, published days after WWDC 2020 in Forbes, by John Koetsier. And, less than a week later, John Koetsier and Eric Seufert, presented a podcast,?Apple’s IDFA is dead. Is this mobile marketing’s apocalypse??Where they discuss developer activities they expect will be negatively impacted by the new?Privacy Rules.

It became clear within a few months the pace of M&A activity was on the rise and Articles such as?Speeding up market consolidation’: Apple’s privacy changes expected to spark wave of gaming and ad tech M&A, written by Lara O’Reilly, published in Digiday, began to appear on a regular basis.

An analysis of M&A activity was completed in July 2021, using data obtained from Crunchbase, regarding “announced deals” for the year following WWDC 2020, to examine the M&A activity of gaming companies.

As the information was analyzed, a hypothesis was further developed to look beyond volume and activity and answer other questions, such as:

• Did the privacy changes impact consolidation activity?
• Was the resulting M&A activity a response driven by the desire, or need, to accumulate more first party data?

The analysis is summarized in?Privacy Island?and examines the four years preceding WWDC and the year immediately following WWDC 2020. The full version is 21 pages and is being updated to include the 2022 and to see if there was a change in activity level.

Consumer

Opt-In Rates:?The implementation of Apple’s

Privacy Changes gave each individual control over who could collect and use their private information. It also created additional steps for the consumer and developer.

The developer must display a message very similar to the one shown at left, if they plan to collect and use information deemed private.

The consumer must complete new actions and navigate new decisions after downloading an app to decide whether or not they want the app to access their personal or device information.

“Compared to early predictions, which placed opt-in rates as low as 5%…we’re seeing numbers well over 30%. Gaming, for example, posted an average of 31% in Q2 of 2022.” —?Adjust Mobile App Trends 2022, A global benchmark of app performance

One thing is clear from these opt-in rates, a large amount of data is no longer available. One reason for this is caused by the double opt-in requirement. Not only must the consumer opt-in to the app they are downloading, they must also opt-in to the app where the ad is displayed, (the publisher app).

Eric Seufert explained that opt-in rates are irrelevant because “publisher opt-in rates are independent of advertiser opt-in rates”. He described this relationship using?Bayes’ Theorem, “where the irrelevance of any given app’s opt-in rate is underscored”.

Source: Adjust Mobile App Trends 2022, A global benchmark of app performance

The opt-in percentages are typically reported from the view of the advertiser. However, when considering the double opt-in requirement, the actual opt-in rates that yield usable data are far lower. With gaming as the example, the graphic above?does not?tell us that 100% of the available data from 31% users is available, it is far less. The available data for each app could be thought of as a block of swiss cheese with holes all over where data used to be.

“The nature of ATT and the mechanics of advertising means that ads targeting and measurement are constrained by the opt-in rates of publishers… A high opt-in rate doesn’t guarantee a similarly high rate of IDFA availability.”

Ad Network(s)

Since May 2018 when GDPR became effective there have been additional privacy changes. None has caused such a rift in the landscape like the one caused by Apples’ introduction of its?Privacy Rules.

“Apple threw the entire mobile ecosystem in a frenzy at WWDC in June 2020 when it announced a radical new change to the way the?Identifier for Advertisers (IDFA) could be accessed and used for advertising tracking.?This change takes the form of the new App Tracking Transparency (ATT) framework: developers will need to collect an explicit opt-in from users before accessing their IDFA. It is believed that very few people will opt into having their IDFAs tracked, which is problematic for advertisers, as the IDFA is the primary means by which iOS advertising campaigns are measured and optimized.” Eric Seufert goes on to say that the IDFA is useful in two ways for advertising purposes:

1. Install Attribution?— which means assigning an install to a specific campaign based on the IDFA of the device that clicked on the ad
2. Event Attribution?— tracking and aggregating subsequent event activity can be attributed to a specific campaign based on the IDFA of the device that installed the app after clicked on the ad

In the absence of the IDFA, an entire ecosystem built around the IDFA would stop functioning until a newer version of the system could be put into place.

As we saw earlier, the opt-in rate wasn’t as bad as originally expected. Still a large amount of data is unavailable, the opt-in rate might be calculated at 31% of games, however, that doesn’t mean that 100% of the available data from 31% of the games is available. It’s far less than that, and the data could be more representative of a block of swiss cheese with holes all over where data used to be.

Proof for these data holes exist, as described in a report earlier this year from Appsflyer, where they state,?“iOS games continue to struggle in aggregate data reality”. As they conduct a year-over-year comparison beginning Q1, 2021 and ending Q1, 2022. In the report they note the following:

• Compared Apps (Gaming and Non Gaming) that were live during the entire period
• Non-Gaming Apps share of organic-driven revenue is much higher, meaning they are less reliant on advertising, and not as impacted as Game Apps.
• IAP Losses for iOS Gaming Apps were double the loss recorded for Android
• Games are heavily reliant on marketing and optimize based on user-level data
• Breaking this addiction to user level data has been hard

Even if an individual opted-in to tracking, Apple’s SKAdNetwork (SKAN) places limits on how much information is sent by Apple to attribute installs and track events. This is because developers receive fewer signals from Apple’s SKAN that make it difficult to track users. In addition, there are additional, compliance-like requirements that ultimately are developer requirements. The developer is responsible for all information collected by their app, even if the information is collected by a third party SDK the developer has integrated and the SDK belongs to an ad network or other service provider. This could make it challenging for developers to report adequately on the data that they collect and expose them to potential risks from the activities of third parties.

Another significant change since the original article is Apples’ intent to get back into advertising after closing iAd in 2016. With the closure of iAd, many developers believed Apple was officially out of the Advertising Business. Developers are severely restricted to the information they can access, collect and use while Apple has the ability to access all of the users’ data and developers are at a disadvantage.

Advertisers — Back to the Future?

The scope of app discovery today requires thinking and acting similarly to the way we thought back in 2013. When the original article was published I recounted walking by a billboard that stated “300,000 apps for everything you love”. Depending on the source you consult, there are as few as 1.7 million apps in iTunes and as many as 5 million.

After the recent Privacy Changes, App Discovery is as broken today as it was 10 years ago.?Nothing has changed, and the “improvements” gained through paid UA have been lost, making the discovery process even more difficult.

In 2013, my co-founder compared the App Stores to Home Depot. They have everything you could possibly want, but finding it is another matter. Finding help can be an exercise in futility.

This isn’t the first time questions have been raised regarding Motivations driving behaviors.?“Taking back control over app distribution…is Apple’s primary motivation in deprecating the IDFA: ads have become the foremost mechanic through which apps are discovered, and Apple has lost total editorial control over content distribution. Related to the point above, Facebook is the primary beneficiary of this dynamic: Facebook ads front-run organic search and discovery, and Facebook effectively serves as the principal point of app discovery.”?—?Eric Benjamin Seufert, Mobile Dev Memo, Why does Apple (finally) care about App Store discovery

Did Apples’ desire for control, or the rapid rise in adoption of smartphone and tablet technology create the problem? Or do the problems with the App Store exist as a result of both factors.

Market Adoption of Smartphones and Tablets

The chart below illustrates the rapid adoption of tablets and smartphones (2010) compared to other technology innovations adopted since 1860.

On the chart above you can see where smartphone and tablet technology start, they are both located to the far right, above the horizontal axis where 2010 is located, and the slope of the line is essentially straight up. Compared to other household technologies, these were adopted in the shortest amount of time with the steepest adoption curve of any other technology. Surely, this had to have contributed in some way to the issues that developed.

Control Leads to Developer Backlash

Was the desire for control embedded in Apple’s DNA? If you consider Apple CEO Steve Jobs’ initial desire to control the apps in the store and not let third party developers build native apps for iOS, instead directing developers to build web applications for the safari web browser.

Developer backlash attributable to the announcement by Jobs, caused Apple to reconsider its position and released an SDK in March 2008, opening the store to third party developers, and changed the traditional software distribution model, and created an opportunity for developers when Apple opened the store for developer submissions in July 2010. The growth that came from this opportunity is another contributing factor to the app discovery problem.

The growth of the App Store has contributed to other developer backlashes. One of the most public was Epic vs Apple. Epic CEO, Tim Sweeney, has been an outspoken critic of Apple on topics ranging from Apple’s 30% charge on In-App Payments to monopolistic and antitrust claims related to Apple. An article written by Malcom Owen does a good job summarizing all related activities through April 4, 2022. In?Epic Games vs Apple trial, verdict, and aftermath — all you need to know.?“While the fight is mostly between Epic Games and Apple, it has already seen other parties wading in with their observations and opinions on the matter, including developers of other apps included in the App Store. Simultaneously, as Apple received scrutiny over its policies, Epic itself has also come under fire for how it handled the situation, including forcing it to happen and orchestrating a premeditated response.”

What’s Next

In the absence of a solution, the market will attempt to create a workable solution to solve the App Discovery problem.

Recapping a few of the insights we gathered regarding App Discovery included:

• People prefer free apps over a paid apps.
• People trust their friends and family's for recommendations
• Individuals want to save time when possible
• The less familiar an app is, the more likely a user is to download it, as a result of a referral

How would you incorporate one or more of these insights into addressing app discovery today?

Over the years, these are the types of analysis and insights that developers are telling us they find beneficial:

Behavioral targeting

• No longer works like it used to
• Data from third parties cannot be joined with data collected from user, unless the user has opted in to collection and tracking
• To be effective, there is a double opt-in requirement
• Companies offering this service are not as viable as they were before Apples’ privacy rules

Contextual Targeting

• Apps that appear most often with other apps
• Apps a developer should target to achieve improved results.
• Identify publishers that yield value greater than others.
• Identify apps that where the predictive value of an install is highest
• predictive value of revenue
• predictive value of retention

If you’d like to see how we’ve used contextual targeting to target without requiring personal or device information, schedule a demo and see how it works for you!

Note: This?article?was originally published in May 2013 and updated to reflect changes over the past 10 years. With the Privacy changes, the information is as relevant today, as it was when it was originally published.